The seismic performance of stone masonry buildings in Faial island and the relevance of implementing effective seismic strengthening policies

Engineering Structures 141 (2017) 41–58

Contents lists available at ScienceDirect

Engineering Structures journal homepage: www.elsevier.com/locate/engstruct

The seismic performance of stone masonry buildings in Faial island and the relevance of implementing effective seismic strengthening policies Rui Maio a,⇑, João M.C. Estêvão b, Tiago M. Ferreira c, Romeu Vicente a a

Department of Civil Engineering of the University of Aveiro, RISCO – Aveiro Research Centre of Risks and Sustainability in Construction, Portugal Department of Civil Engineering of the University of Algarve, Portugal c Department of Civil Engineering, University of Minho, ISISE – Institute for Sustainability and Innovation in Structural Engineering, Portugal b

a r t i c l e

i n f o

Article history: Received 7 July 2016 Revised 2 March 2017 Accepted 6 March 2017

Keywords: Stone masonry buildings Seismic performance Seismic retrofitting Tremuri software Non-linear static (pushover analysis) Fragility curves

a b s t r a c t Enhancing the seismic performance of traditional stone masonry buildings is considered a crucial measure towards the preservation and safeguarding of our built heritage, particularly in seismic prone areas such as the Azores archipelago, in Portugal. In this context, the present paper is focused on the seismic vulnerability assessment of two traditional stone masonry building located in the Faial Island, taken as typologically representative of the rural Azorean building stock. The case study buildings were modelled based on the equivalent frame model approach and non-linear static (pushover) analyses were performed to assess their seismic performance. Firstly, results were compared in terms of capacity curves, and secondly, two different seismic performance-based assessment methods (N2 and CSM) were used to determine the respective performance points and assess the seismic safety of both structures. The seismic demand was defined by a set of real ground motion records, which the authors assumed as representative of the 1998 Azores earthquake, both in terms of magnitude and epicentral distance. Additionally, a set of traditional retrofitting solutions were implemented to the original models in order to analyse and compare their influence over the building’s global seismic performance. These solutions were adopted in the aftermath of the 1980 and 1998 Azores earthquakes by different design offices and engineering consultants, based on the designing recommendations specifically prepared by the Regional Laboratory of Civil Engineering (LREC) under the scope of the Faial rehabilitation process. Finally, fragility curves were derived based on the spectral response approach, and the results were critically discussed. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction 1.1. Seismic risk in Portugal Earthquakes are one of the most frightening, destructive and deadliest natural disasters. In the European context, Portugal represents an important seismic prone area, as the latest estimations indicate that within the next 50 years this country incurs the risk of being severely hit by a strong earthquake similar to the historical 1755 Lisbon earthquake, in which more than 50% of the building stock is expected to undergo heavy levels of damage or even destruction, and about 10% of the population of Lisbon is expected to perish [1]. On the one hand, mainland Portugal is located in the southwest part of the Eurasian plate, near the southern border of

⇑ Corresponding author. E-mail address: [email protected] (R. Maio). http://dx.doi.org/10.1016/j.engstruct.2017.03.009 0141-0296/Ó 2017 Elsevier Ltd. All rights reserved.

the African and North-American plates, being subjected to both offshore and onshore seismic activity with large to very large and moderate to large magnitudes, respectively [2]. On the other hand, the Azores Archipelago, located at the triple intersection of the Eurasian, North-American and African (Nubian) plates, allies its volcanic origin with important tectonic activity, being therefore considered the most hazardous region of Portugal [3]. The seismic background of Portugal is beyond doubt remarkable. Among other minor events and apart from the well-known 1755 Lisbon earthquake, in 1909, a moderate event of magnitude 6.0 M w struck the village of Benavente [4], causing 46 fatalities and damaging more than 3000 buildings [5]. More recently, in 1998, an offshore event of magnitude 5.8 M L [6] struck the Islands of Faial, Pico and São Jorge, destroying roughly 70% of the building stock and causing 8 fatalities, over than 100 injured, and many thousands of homeless [7]. This earthquake has made possible to gather an unprecedented amount of data concerning the

42

R. Maio et al. / Engineering Structures 141 (2017) 41–58

characterisation of the building stock and respective damage mechanisms [8,9]. It is worth referring that both case studies herein analysed were identified and selected by accessing this database. One of the most widely used definitions of risk was given by Cardona and Barbat in [10], and defines risk RiejT , as the probability of loss in an exposed element e as a consequence of the occurrence of an event with intensity larger than or equal to i during an exposition period T. According to these authors, it can be represented by the Eq. (1), where the function, f, is the mathematical convolution between hazard, Hi , and vulnerability, V e , during an exposition period, T.

RiejT ¼ f ðHi  V e ÞjT

ð1Þ

Several studies have been developed during last decades, either addressing the evaluation of the seismic risk in Portugal or focusing on a particular aspect of its definition. In Silva et al. [1,11] for example, the authors revisited most of the hazard-related methods that have contributed to the understanding of the Portuguese seismic hazard and risk. In Maio et al. [12,13] instead, the review of some of the most widely used vulnerability-related methods was carried out. As discussed in these noteworthy references and highlighted in Santos et al. [14], such seismic risk analyses must be necessarily based on an in-deep knowledge of the seismic response of samples of buildings considered to be representative of a particular structural system or a given class of buildings and, in this sense, the analysis of typologically representative buildings that can serve as basis for large-scale assessment methodologies is a clear and continuing need.

Tagus Valley and Algarve regions in mainland Portugal, this is seen as an utterly lack of sense of opportunity, knowing moreover that the cost associated with most of these traditional seismic strengthening techniques have little impact over the total cost of renovation when integrated in the structural project design. Against this drawback, the development of a reliable and accurate methodology for the seismic vulnerability assessment of the existing old masonry building stock is seen as a crucial step towards the seismic risk mitigation in historical urban areas. According to Silva et al. [11], such a methodology should include the prioritisation of regions where both retrofitting and strengthening campaigns should take place, the creation of insurance and reinsurance schemes to transfer and share the consequent financial burden between governments and private sector, planning and managing emergency response at an urban or regional scale and the definition of regulations to endorse seismic-resistant construction practices. The seismic risk mitigation of the building stock inserted in these particular areas will not only allow us to guarantee an appropriate safety level for local communities but also to pass on our built heritage to future generations. Nevertheless, it is outside urban areas’ boundaries that higher levels of destruction are often found, not only due to poorer quality of construction materials and detailing, but also to poorer soil and foundation quality, as observed in recent earthquakes such as July 1998 earthquake in Azores, the April 2009 earthquake in L’Aquila or more recently, the April 2015 earthquake in Nepal [18–20]. Bearing in mind this framework, the authors also aimed to balance the attention given in the past few years to the conservation of our built heritage within seismic prone areas, by focusing on the seismic assessment of traditional stone masonry buildings located in a rural environment.

1.2. Existing masonry building stock Despite all the huge technological advances, innovative materials and building systems that we have been recently witnessing, load-bearing masonry construction still represent one of the most widely used building systems worldwide. Furthermore, the representativeness of such building system when assessing old urban centres is even more relevant. It is estimated that this building system, typically more vulnerable to earthquakes and therefore at significant risk even when only subjected to moderate events, still represents more than half of the Portuguese building stock [15]. Nonetheless, the seismic performance of such structures differs substantially as a function of multiple factors such as architectural configuration, structural details, quality of materials and execution works, the presence of inappropriate or weakening interventions, or even the boundary conditions and interaction between adjacent buildings [13]. At a time when the crisis in the Portuguese construction sector seems to have been overcome as a result of a paradigmatic shift towards the renovation and rehabilitation of existing building stock, large-scale investments have been gradually made in focal historical centres across the nation. However, two major issues of great concern for the scientific community in general have arisen. Firstly, as the economic attractiveness seems to be the main if not the only interest behind the preservation of urban cultural heritage, the authors fear for the risk of excessive ‘‘façadism”, a phenomenon that together with low-cost renovation strategies is threatening the adequate preservation of both tangible and intangible values of our cultural heritage. Secondly, the approval of the Decree-Law No. 53/2014 that rules the renovation of existing buildings [16] is indeed the most worrisome aspect because it calls into question the sustainability of such interventions, as they usually do not respect basic recommendations for seismic design neither consider the adoption of traditional and cost-effective strengthening techniques [17]. In seismic-prone areas such as the

2. Traditional basalt stone masonry buildings from the Island of Faial The earthquake that struck the Island of Faial, Pico and S. Jorge on the 9th of July 1998 allowed the collection of an unprecedented quantity of good quality data on damage in constructions. The information collected during the 10-year reconstruction process of Faial Island, conducted by the Society of Promotion for Housing and Infrastructures Rehabilitation (SPRHI), was gathered over an 8month period in 2007 by the Regional Secretariat for Housing and Equipment (SRHE) of Faial and assembled in the book ‘‘Sismo 1998 Açores – uma década depois” edited by Oliveira et al. [7]. The quality and uniqueness of this database in both national and international context have encouraged the development of several advanced studies throughout the following years. Even though the full access to the mentioned database was made available, complementary field work, prospected in the scope of the FCT URBSIS project – Assessing Vulnerability and Managing Earthquake Risk at the Urban Scale - was carried out by the respective research team in order to understand the evolution and diachronic process resulting from rehabilitation interventions implemented since 1998. More recently, Neves et al. [9] carried out a comprehensive characterisation of the building stock of the Faial Island, describing the most common structural typologies, contributing for the development of the referred database. According to the authors, at the time of the earthquake, stone masonry buildings represented about 60% of Faial’s building stock (estimated from a total sample of 2305 buildings). Despite several factors that have been previously pointed out as influencing the seismic response of masonry buildings, with this study, the authors aimed to draw attention to the seismic response and vulnerability of traditional stone masonry buildings within rural areas of Faial Island, a topic discussed in the first place by Costa in [21]. Throughout research

R. Maio et al. / Engineering Structures 141 (2017) 41–58

and in situ inspection carried out in Faial Island, it was perceived that traditional rural stone masonry buildings can be broadly divided in one-storey buildings, generally modest and located in flat rural areas, and two-storeys buildings, most common in urban centres of rural towns, terraced, or detached when in slopes. Moreover, these buildings are mostly isolated structures, and therefore direct interactions with adjacent buildings, an aspect of utter relevance when assessing the seismic vulnerability of buildings in aggregate usually observed in urban areas, are in this particular case inexistent. Regarding their structural system, these two building types generally present external basalt stone masonry walls as the main load-bearing elements. Depending on the economic resources of the owners, two different types of construction typologies can be clearly identified in traditional rural Azorean constructions: regular masonry walls made of well-cut and shaped stone blocks; and irregular masonry walls with ‘‘burnt stone” blocks, arranged and treated with more or less care depending on the master builder [22]. In addition to [7,21], further literature concerning the seismic vulnerability assessment of stone masonry typologies in Faial island can be found in the studies of Costa and Arêde [23], Costa et al. [24], and Costa [25], for example. Moreover, several other studies focusing the seismic vulnerability of the old city centre of Horta can be found for instance in [7,9,26–28]. As previously mentioned, this study addresses the seismic performance of two traditional stone masonry buildings located in the rural parish of Cedros, taken as representative of the Azorean rural building typology. Within the rural built environment, these case studies were insightfully selected to cover both regular and irregular typologies, as explained in the following paragraphs. The required information related to these buildings was accessed from the respective reports made available by SRHE, which allowed the authors to fully understand and describe the building geometry, structural system and material properties. As described in the respective reports, the vertical structure of the two-storeys isolated rubble stone masonry buildings, herein used as case studies and illustrated in Fig. 1, are composed of basalt, cinerite, andesite and other small-size stones rendered with mortar. While the load-bearing walls’ thickness is about 0.7 m thick for both buildings, their plan configuration varies considerably. While Building A, in Fig. 1 (left), presents a rectangular-shaped (regular) plan of 9.0 m by 8.0 m and a total height of 5.3 m, Building B, in Fig. 1 (right) presents an L-shaped (irregular) plan with 16.0 m by 13.0 m as larger dimensions, and a total height of 5.1 m. Another particularity of this building typology, is that the access to the upper floor is traditionally made through an exterior stone masonry staircase, which represents an important mass and therefore should not be neglected when assessing the seismic performance of the structure due to the resulting torsion effects. In the case of Building B, the structure of the exterior staircase is com-

43

posed of reinforced concrete elements (columns, beams and voided slab), as shown in Fig. 1 (right). With respect to the characteristics of horizontal diaphragms, and even though detailed information and measurements were not available for these particular case studies, through the in situ inspection and survey campaign carried out by Ferreira et al. [28], it was possible to confirm the predominance of flexible timber floors within the building stock of Faial Island, most of which have been subjected to deterioration and ageing phenomena over time. Thus, the traditional floor structure is composed of timber planks supported by timber joists. The roof structure consists of a traditional kingpost hip roof truss of four hip rafters. Finally, foundations are usually composed of stone masonry footing of about 1.0 m deep for two-storeys buildings, erected using the same stone masonry typology but slightly thicker than the load-bearing walls. 3. Seismic in-plane behaviour of stone masonry buildings The current section presents the properties and assumptions adopted in the numerical models, for both case study buildings’ original condition, i.e. before retrofitting (hereinafter referred as BR condition), and the respective results obtained from the seismic performance-based assessment carried out. 3.1. Seismic response of the BR original condition In order to assess the seismic global performance of these stone masonry buildings briefly described in the previous section, a three-dimensional model was initially developed resorting to 3MuriÒ software commercial version (release 10.5.0.4) [29], and subsequently non-linear static (pushover) analyses were performed by using the research version of the Tremuri program [30,31], which is based on the equivalent frame model (EFM) approach, following the assumption that the in-plane response of masonry walls with openings may be discretised by a set of macroelements, illustrated in Fig. 2. While piers are vertical elements that support both dead and live loads (in orange on the numerical mesh presented in Fig. 2), spandrels are horizontal elements placed in between two vertically aligned openings, which couple piers in the case of seismic loads (in green). Finally, rigid nodes (in light blue) consist of undamaged masonry portions confined between piers and spandrels. The strength criteria governing the in-plane behaviour of piers and spandrels was defined according to both EN 1998-1 [32] and the Italian Code for Structural Design – NTC [33] and modelled through non-linear beams [30]. The flexural response, combining compressive and bending failure, is based on the beam theory, neglecting the tensile strength of the material and assuming a rectangular normal stress distribution at the compressed toe [34]. The shear response is governed by the

Fig. 1. Front view of the case study Building A (left) and Building B (right), both located in the parish of Cedros, Faial Island.

44

R. Maio et al. / Engineering Structures 141 (2017) 41–58

Fig. 2. Three-dimensional view of the case study Building A (left) and Building B (right) located in the parish of Cedros, Faial Island and the corresponding macroelements of the mesh generated for each façade wall: piers, spandrels (or lintels) and rigid nodes, respectively in orange, green and light blue. Please note that façade walls are aligned with the U x direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

diagonal cracking failure criteria, originally developed by Turnšek and Sheppard [35], and recently adapted for existing masonry buildings by the NTC [33]. Despite some unavoidable approximations of the actual structural behaviour, mainly concerning the mechanical description of damage and dissipation mechanisms, this simplified formulation enables, amongst other features, to perform pushover analyses with a very reasonable computational effort [36]. Determining masonry walls’ mechanical properties when adopting numerical approaches is essential towards the robustness of the subsequent models, as properties such as the Young’s modulus, for example, has great influence over the seismic response of traditional stone masonry buildings, and therefore, over the design of rehabilitation and strengthening projects. However, with the exception of few experimental works, such as the in situ testing campaign carried out by Costa [21] or Ferreira et al. [22], there is a general lack of information and research related to the characterisation of basalt masonry walls’ typical of the Azorean Islands. Even though the value suggested in [21] for the mass density, c, of these irregular basalt masonry walls is quite similar to the minimum value proposed in the NTC [33] for a similar stone masonry typology – ‘‘masonry in disorganised (irregular) stone” - the value obtained for the Young’s modulus is significantly lower (200 MPa), when compared to the minimum value recommended in the NTC 2008 [33] for the same stone masonry typology (690 MPa). However, when applying the reduction factor equal to 0.5, after [36], recommended both in the EN 1998-1 [32] and NTC [33] to consider a cracked condition in the stiffness properties of existing masonry panels (E and G), the difference between the values of the elastic properties suggested in [21] and the minimum values given in the NTC 2008 [33] for the ‘‘masonry in disorganised

stone” equivalent typology is more blurred. Therefore, in order to avoid eventual convergence problems and to be more consistent in all of the input values required to characterise the loadbearing walls, the authors have considered the minimum values suggested in the NTC 2008 [33] for the mentioned typology ‘‘masonry in disorganised stone”, considering a cracked condition in the stiffness properties of existing masonry panels and a knowledge level KL1, to which corresponds a confidence factor CF KL1 equal to 1.35 [32]. Hence, mass density, c, was assumed equal to 19 kN/m3, the Young’s modulus, E, equal to 690 MPa, the Shear modulus, G, equal to 230 MPa, compressive strength of masonry, f m , equal to 60 MPa, and ultimate shear strength, s0 , equal to 2 MPa. When using 3MuriÒ, floors and roofs can be defined as orthotropic membrane finite elements [30], characterised by an equivalent thickness, s, a Young modulus E1;eq and E2;eq , respectively in the orthogonal and perpendicular direction of the floor warping, and an equivalent shear modulus, Geq . Due to the lack of information regarding the mechanical properties of horizontal diaphragms, the shear stiffness values suggested either by Giongo et al. [37] and the NZSEE guidelines [38], after [39], for a straight sheathed flexible timber floor were adopted for a poor rating condition. Thus, a single straight sheathing membrane with an the equivalent shear modulus, Geq , equal to 6.8 and 9.0 MPa, was respectively adopted for the case in which the direction of loading is parallel and perpendicular to timber joists (with Young modulus E1;eq equal to 7 GPa and cross section 10 cm2x10 cm2 spanned in 40 cm). Moreover, timber joists were considered to be poorly connected to the membrane and to the perimeter walls. Finally, gravity (Gk ) and live (Q k ) loads equal to 1.00 and 2.00 kN/m2, respectively, were assigned to timber floors, according to [40]. To the timber roofs, the

R. Maio et al. / Engineering Structures 141 (2017) 41–58

values of 1.00 and 0.30 kN/m2 were assigned for Gk and Q k , respectively. It is worth noting that while the vertical structural system of Building A is only composed of stone masonry with the abovementioned mechanical properties, in Building B, the structure of the exterior staircase and adjacent balcony is made of reinforced concrete. These elements were defined with a cross-section 30 cm2  20 cm2 with 4U16 longitudinal steel bars, concrete cover of 3 cm, 6U stirrups spacing 30 and 15 cm, respectively to R.C. beams and columns. Again, due to the lack of information regarding the mechanical properties of reinforced concrete elements, a class C16/20 and B420 was assigned to concrete and steel rebars, respectively, by naturally assuming a knowledge level KL1. These classes were assigned assuming these elements as poorly executed and made of low quality materials. Moreover, acknowledging the important contribution that exterior stone masonry staircases have on the global seismic response of masonry buildings, and facing the limitation of 3MuriÒ software on modelling such elements, they were herein represented by means of pier macro-elements with the same mechanical properties as load-bearing walls. While reinforced concrete balconies were assigned with Gk and Q k values equal to 3.00 and 4.00 kN/m2, respectively, to stone masonry balconies the values of 0.70 and 4.00 kN/m2 were assumed [40].

3.2. Non-linear static (pushover) analyses There are different options available in the literature for the choice of the mathematical modelling and type of analysis for the structural assessment of existing structures [41]. The uncertainties related to these models are directly reliant on the level of complexity desired to conduct the fragility assessment. In this study, the in-plane capacity of these models were obtained by means of non-linear static (pushover) analyses, exposing the structure to a static lateral load pattern. These analyses were performed following the recommendations of EN 1998-1 [32] and the NTC [33], i.e., by considering each main direction (U x and U y ), two different load pattern distributions (uniform, proportional to mass, and pseudo-triangular, proportional to the product between mass and height) and also accidental eccentricity of the centre of mass with respect to the centre of rigidity (5% of the maximum planar width perpendicular to each seismic action direction). Unlike Finite Element based programs, Tremuri program does not allows the user neither to select nor create a fictional control node located in the centre of gravity of the roof level of each structure, as suggested in the Eurocode. It is worth noting that pushover curves were plotted until reaching a 20% decay of the maximum base shear capacity, V b , again according to the referred codes, which relates this condition to the ultimate displacement capacity of the structure, and by considering the average displacement of all nodes located at the roof level, dn . Despite the latter consideration, that allows obtaining a more representative curve of the global behaviour of structures with flexible diaphragms, rather than the behaviour of a particular wall, the execution of pushover analysis requires choosing a single control node. Therefore, to overcome this issue, the authors adopted a criterion based on the evaluation of the ratio between the maximum admissible ground acceleration, ag;max , compatible with the fulfilment of the ultimate limit state (taking into account q < 3), and the reference ground acceleration, agR (considered equal to 2.50 m/s2), for all the perimetral nodes located at the upper storey (identified in the previous Fig. 2). Thus, a sensitivity analysis was carried out to understand which control node provided the lowest ag;max =agR ratio (nodes N3 and N42 in the case of Building A, and nodes N22 and N6 in the case of Building B, respectively in the U x and U y directions. Finally, the selection of these control nodes was validated by observing the damage distri-

45

bution for each pushover analysis, from which it was possible to perceive that the macroelements near to the selected control node are damaged for just a few steps, as recommended from the analytical viewpoint (see Fig. 4). In the following Fig. 3, the pushover curves concerning to the most unfavourable analyses selected according to the abovementioned criterion were plotted for each main direction and type of load pattern distribution. According to Fig. 3(a), which refers to Building A, it is possible to observe that both the stiffness and base shear capacity, V b , are higher in the U x direction, motivated by a higher cross-section and subsequently resistant area in this direction and also due to smaller area of openings. Moreover, unlike what was observed in the U y direction, the available ductility in the U x direction is higher in the curve corresponding to the pseudo-triangular distribution. The Fig. 3(b) instead shows that in general terms the Building B presents higher shear base capacity than Building A. In this case is worth referring that in both directions the uniform pattern distribution gives higher capacity and ultimate displacements. Notwithstanding the previous discussion, Simões et al. [42] have argued that in case of very flexible floors, the adoption of the referred criteria governing the stop-condition of pushover curves, proposed by the EN 1998-1 [32], may lead to nonconservative results, as different walls behave almost independently, meaning that a very moderate redistribution of seismic loads occurs among masonry walls. Hence, severe damage in walls may not appear evident on the structure capacity curve, which is representative of the structure’s global behaviour. In this sense, it is recommended the assessment of the damage pattern distribution obtained for the last steps of the most unfavourable analyses. In this sense, Fig. 4 presents the damage distributions obtained in the walls where the control node is located, which corroborates the explanation given above, concerning the early damage of the macroelements located near the selected node control. From this figure it is possible to observe that shear damage mechanism prevails in both buildings, presenting shear cracking in the upper storey and shear failure in piers of the first storey. Despite no information regarding the real cracking pattern distribution of Building B was available, in the case of Building A, the results obtained are not entirely consistent with the real cracking pattern distribution observed in the aftermath of the 1998 Azores earthquake (see Fig. 5), from where it is evident a global bending damage pattern in piers and spandrels. Several reasons might justified this inconsistency, such as the variability between the real ground motion and the elastic response spectrum designed according to the EN 1998-3 [43], the occurrence of mixed mechanisms (in- and out-of-plane) which are not captured in the Tremuri program, or even the non consideration of external masonry staircases’ influence due to modelling limitations.

3.3. Seismic performance-based assessment In the last decade, the use of non-linear static procedures for the assessment of the seismic performance of buildings have become more and more attractive from the user standpoint [36]. Essential elements of a performance-based assessment procedure are demand and capacity. While demand is a representation of the earthquake ground motion, capacity is a representation of the structure’s ability to resist the seismic demand. The seismic performance-based assessment of both case studies is herein carried out according to two different simplified methods, referred to as non-linear static analysis procedures that are commonly used  to determine the structure’s performance point, dt , which is in turn computed from the intersection between the capacity curve of the structure and the seismic demand in terms of response spectrum.

46

R. Maio et al. / Engineering Structures 141 (2017) 41–58

(a) Building A

(b) Building B

Fig. 3. Pushover curves representing the base shear, V b , against the respective control node displacement, for the case study Building A (a) and Building B (b) modelled with 3MuriÒ for the BR (before retrofitting) condition.

Fig. 4. Damage pattern distribution on each structural macroelement of Building A (on top) and Building B (on bottom), for the most unfavourable control node and respective pushover analysis (in the U x and U y directions, respectively). Please note that the wall scale deformation factor was set equal to 100 to better display the damage mechanisms on each macroelement.

Fig. 5. Real cracking pattern distribution observed in Building A in the aftermath of the 1998 Azores earthquake.

These are the N2 Method and Capacity Spectrum Method (CSM), respectively recommended by the EN 1998–1 and the ATC-40 for deriving the capacity curve and subsequently the performance

point of a given structure. With this exercise, the authors aim to compare the results obtained by applying these two different methods, which mainly differ on the definition of the seismic

47

R. Maio et al. / Engineering Structures 141 (2017) 41–58

demand and whose main features are described in the following paragraphs. Therefore, the ultimate goal of this comparison is to understand the variability of the seismic demand in common outputs such as fragility curves. 3.3.1. N2 Method The capacity curves presented in the following Fig. 6, were obtained by converting the pushover curve from the original Multi Degree of Freedom (MDoF) system (in the previous Fig. 3)) into an equivalent Single Degree of Freedom (SDoF) system [34], according to the N2 Method, originally proposed by Fajfar [44] and adopted on the current EN 1998-1 [32] and NTC [33]. According to the referred codes, an elasto-perfectly plastic force–displacement relationship was assumed to define the SDoF capacity curve, wherein the initial stiffness was determined based on the intersection with the point corresponding to 70% of the maximum base shear strength achieved on the initial branch of the pushover curve. Moreover, the yielding strength, F y , was determined so that the areas under the pushover curve and the elasto-perfectly plastic capacity curve are equal. Similarly to what was observed in the pushover curve, Fig. 6(a) shows that Building A presents a higher limited structural strength but lower ductility in the U x direction. In the case of Building B, a huge discrepancy between the two load pattern distribution types was observed both in terms of limited structural strength and available ductility. Even though the difference between the two main directions is more tenuous, it is possible to observe that, for the worst case scenario analyses, Building B presents higher strength in the U x direction. The verification of the ultimate limit state (ULS) consisted on checking if the structure withstands the seismic demand, which was defined in this case by means of an elastic response spectrum with a return period, T R of 475 years, according to the recommendations of the EN 1998-1 [32], the EN 1998-3 [43], and the NP EN 1998-1:2010, when using the recommended parameters established in the Portuguese National Annex for the Azores region [45]. Moreover, an equivalent viscous damping n, of 5% for a foundation soil type C was adopted in both cases. The buildings’ importance factor cI , was considered equal to 1. The seismic performance-based assessment was then conducted by evaluating the following indicators, recommended both by EN 1998-3 [43] and the NTC [33] for the ULS safety verification. The latter recommends the ratio q , computed between the acceleration in the structure with unlimited elastic behaviour, SeðT  Þ, and limited structural strength F y =m , to be lower than 3, aiming

(a) Building A

to limit the overall acceptable ductility of the building. Moreover, according to these same codes, safety is verified when the ratio   between ultimate and target displacements du =dt is larger than 1. Finally, in which regards the ratio between the maximum admissible ground acceleration, ag;max , compatible with the fulfilment of the ultimate limit state (taking into account q < 3), and the reference ground acceleration, agR (considered equal to 2.50 m/s2), safety is verified if this ratio is larger than 1. Table 1 summarises the most relevant parameters of the most unfavourable analyses for each main direction (U x and U y ) and lateral load pattern distributions (uniform and pseudo-triangular, hereinafter referred as triangular). In terms of equivalent period T  , the differences between the main directions in the case of Building A suggests a higher deformability in the U y direction, characterised by a higher area of openings, consistent with pushover curves. With respect to ductility l , which is given by the ratio   between ultimate and yielding displacements du and dy , respectively, higher deviations were found again in Building A. The value obtained for the limited structural strength F y =m in the case of Building A U x direction, is almost three times higher than the corresponding value obtained for U y direction, again justified by the number of openings and effective section of structural walls. Excepting the U x direction of Building B, these results obtained from the application of the N2 Method procedure show that practically all analyses failed to verify the previously-mentioned safety requirements, which has driven the need for seismic retroffiting, detailed in the next Section 4. 3.3.2. Capacity Spectrum Method The Capacity Spectrum Method (CSM) instead, is based on the intersection of the capacity curve and a reduced response spectrum to estimate the maximum displacement. Widely applied to assess the seismic performance of these structures [46] when subjected to recorded earthquakes, this method requires the definition of a bilinear representation of the capacity spectrum to estimate the effective damping and appropriate reduction of spectral demand, as described in [46]. In contrast to the N2 Method, the verification of the ultimate limit state (ULS) consists on the assessment if the structure withstands the seismic demand, which was herein defined by a set of 6 real ground motion records believed to be representative of the 1998 Azores earthquake, both in terms of magnitude, ML , and epicentral distance, D (see Table 2), instead of using the elastic response spectrum as in the previous Section 3.3.1. It is worth referring that Rec. 6 was the only real ground motion that

(b) Building B

Fig. 6. Bilinear equivalent capacity curves of Building A (a) and Building B (b) corresponding to the worst case scenario analyses in the BR (before retrofitting) condition.

48

R. Maio et al. / Engineering Structures 141 (2017) 41–58

Table 1 Main parameters of the most unfavourable analyses (BR original condition) resulting from the application of the N2 Method procedure. Building

Control

Direction

Ecc.

node A

N3 N42

B

N22 N6

Load

C

T

l

F y /m

q

pattern

[–]

[s]

[–]

[g]

[–]

du /dt [–]

[–]





ag;max /agR

U x U x þU y þU y

+  + +

Uniform Triangular Uniform Triangular

1.28 1.28 1.36 1.36

0.18 0.22 0.23 0.27

1.21 4.83 1.58 1.60

1.97 0.95 1.63 0.93

4.13 8.52 4.98 8.20

0.23 0.53 0.30 0.19

0.28 0.53 0.31 0.19

þU x U x þU y þU y

  + 0

Uniform Triangular Uniform Triangular

1.03 1.03 1.29 1.29

0.22 0.25 0.19 0.21

3.75 3.07 3.25 3.27

4.58 4.05 0.95 0.77

1.77 2.00 8.56 10.57

2.00 1.52 0.30 0.27

1.94 1.52 0.32 0.28

Table 2 Real ground motion records considered in this study. Earthquake Rec. Rec. Rec. Rec. Rec. Rec.

1 2 3 4 5 6

Sierra Madre Sierra Madre Chalfant Valley 86C Mt. Lewis Livermore Azores

Country U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Portugal

Date 28 June 1991 28 June 1991 30 July 1986 31 March 1986 26 January 1980 9 July 1998

was actually recorded during the 1998 Azores earthquake, which was obtained in the SMA-1 station, located at the Observatório Príncipe do Mónaco, in Horta (Faial Island). Apart from being considered fairly incomplete [47], another issue with this record (identified as Rec. 6 in Table 2) is related to the characteristics of the station’s location, which has been proven to generate very high site amplification’s effect, namely during the previous 1980 Azores Earthquake [48]. To overcome this issue, five other real ground motion records (identified as Rec. 1 to 5 in Table 2) were selected from the database of the Center for Engineering Strong Motion Data (CESMD), corresponding to U.S.A. recorded earthquakes with the same magnitude (M L = 5.8) of the 1998 Azores Earthquake [6]. Since the distances between these earthquake’s epicenter and Faial Island’s shoreline are in between 7 and 25 km, the remaining records were selected from the CESMD database within this range of epicentral distances (see Table 2). The first record (Rec. 1) was established as a baseline reference for the ground motion for the closest distance shoreline, because it corresponds to a rock outcropping site at an approximate epicentral distance of 7 km. The remaining four records were obtained at sites with shear velocities closer to the upper limit of the EN 1998-1 foundation soil

Depth [km]

Station

12.0 12.0 9.0 6.0 7.3 2 to 5

CGS–CSMIP 24399 CGS-CSMIP 24401 CGS-CSMIP 54171 CGS-CSMIP 57191 CGS-CSMIP 57528 Horta Observatory (SMA-1)

ML [–]

v s;30

[s1]

D [km]

5.8 5.8 5.8 5.8 5.8 5.8

680 379 303 282 383 (?)

6.7 20.2 13.6 14.4 6.6 14.0

type C. Only the two horizontal vibration components were considered, making a total of 12 accelerograms. plus the resulting mean response spectra (shown in Fig. 7). Please note that the elastic response spectra, EC8 (in red), was obtained according to the EN1998-1 and EN 1998-3 recommendations. Due to the lack of detail information regarding the v s;30 value on the exact site in which these two case studies were located (Cedros parish), the authors assumed the same dynamic characteristics of surface layers of the nearby city of Horta, based on the study developed in [49]. From this study a v s;30 value equal to 361 m s1 was estimated, to which corresponds the transition zone between soil type C and B. Hence, since the superficial layer is a class D soil (v s;30 lower than 180 m s1) the authors believe that a class C can be considered well adjusted for the present case study buildings. Finally, target displacements (performance points) were determined using a computer implementation of the ATC-40 method [46], using the same set of computer routines that were adopted in other earlier studies developed in the Azores archipelago [50].   For all the cases resulting in structural collapse (du < dt ), the target  displacement, dt , was defined as the point corresponding to the

Fig. 7. Response spectra in terms of period T (left) and spectral displacements Sde (right).

R. Maio et al. / Engineering Structures 141 (2017) 41–58

maximum non-linear damping capacity of the structure. These results are presented further on in Section 5.1, together with those obtained for the retrofitting packages presented in the next Section 4.1. 4. Seismic retroffiting of stone masonry buildings As the massive demolition and replacement of these structures seems neither affordable nor feasible due to historical, cultural, social and economic constraints, this section describes in detail the retrofitting strategies adopted in this study, which are based on the reconstruction methodology defined shortly after the 1998 earthquake by the Regional government of Azores, aiming to enhance the seismic performance of existing stone masonry buildings, through traditional retrofitting solutions compliant with the mentioned constraints. Some of these traditional solutions are still in use these days, as demonstrated for example in [51]. 4.1. Numerical application of traditional retrofitting solutions Bearing in mind the above, six retrofitting solutions of increasing intrusiveness and cost (from S1 to S6), grouped into three cumulative retrofitting packages (from RP 1 to RP3 ), were herein addressed (see Table 3). Despite latest developments on innovative steel connections for the retrofit of timber floors in ancient masonry buildings [52–55], in this study the authors opted by considering the same solutions that were adopted in the aftermath of the 1980 and 1998 Azores earthquakes by different design offices based on the design recommendations specially prepared in the Faial rehabilitation process [56], which were developed by the Civil Engineering Regional Laboratory of Azores (LREC) in partnership with several experienced engineers and consultants in this field [7,57,58]. The comparison between the seismic performance of the original BR condition of the assessed structure and each one of these retrofitting packages is going to be discussed in the following section. As the adopted strategy pursues the cumulative implementation of retrofitting solutions and the authors have considered a first set of retrofitting solutions effective on enhancing the box-like behaviour of stone masonry buildings, resorting to low-tomoderate intrusiveness, the following solutions S1 to S4 were grouped in RP 1 package, as shown in Fig. 8. The retrofitting of wall-to-wall connections by means of effectively tying walls together with steel tie-rods, addressed in the retrofitting solution S1 , is an ancient provision to enhance the building integrity, seen as a crucial requirement to prevent out-of-plane collapse during an earthquake, which has been used for centuries in many Mediterranean European countries. These threaded steel tie-rods 16 mm thick are usually installed horizontally beneath floors (S1 ) and roofs (S4 ) on both sides of the wall, and restrained Table 3 Seismic retrofitting solutions adopted for traditional stone masonry buildings. Retrofitting package

Retrofitting solution

Description

RP1

S1

Wall-to-wall connection improvement through tie-rods Floors stiffening with diagonal bracing and new timber planks Wall-to-floor connection improvement Wall-to-roof connection improvement through tie-rods Wall-to-roof connection improvement through concrete strapping beams Stone masonry consolidation through reinforced plaster

S2 S3 S4 RP2

S5

RP3

S6

49

at the ends by steel anchor plates, as depicted in Fig. 8(a), adapted from Carvalho et al. [56] and D’Ayala and Speranza [59]. This solution is not only effective in increasing the stiffness of flexible floor diaphragms but also in enhancing the connections with exterior load-bearing walls and frontal walls. These tie-rods were implemented in the models of both case study buildings as illustrated in the following Fig. 9. Furthermore replacing all deteriorated structural timber elements of horizontal diaphragms by new elements adequately connected, restoring their original resistant capacity, the solution adopted in this study for retrofitting of floors and roof connections S2 , joins two different stiffening provisions: the installation of 75 mm thick diagonal timber braces at both floor and roof level between timber joists, anchored with /10 galvanised steel threaded rods and 3.0 mm thick galvanised steel angle brackets as well as a new layer of timber sheathing, laid perpendicular to the existing one and adequately nailed to the floor as shown in Fig. 8(b). However, given the Tremuri program limitation on designing such a particular provision as diagonal bracing, the authors have opted instead to sustain their reasoning based on the NZSSE guidelines, after ASCE (2014), that proposed stiffness multipliers for other forms of flexible timber diaphragms. Therefore, the authors consider a double straight sheathing (with diagonal sheathing chorded) membrane typology using a fair rating condition to the new timber floor provision included in this retrofitting solution, S2 . Hence, a multiplier of 9.0 was applied to the Gd values of 285 and 215 (in MPa), respectively for the load direction acting parallel and perpendicular to timber joists. The equivalent shear modulus Geq was then obtained by dividing Gd by the thickness of the membrane (double straight sheathing), which was considered equal to 4.5 cm (2.5 cm related to the original single straight sheathing typology assigned in the BR condition plus 2.0 cm of the new timber sheathing layer). The retrofitting of wall-to-floor connections solution S3 was enhanced by introducing 3 mm thick full-length steel angle brackets adequately anchored to walls through steel connectors and anchor plates, as depicted in Fig. 8(c), complementing the previous solution S2 . However, again due to modelling limitations, retrofitting solution S3 was implemented in the model by assuming timber joists with square cross-section 10 cm large and spaced by 40 cm. Moreover, these elements were considered well connected to load-bearing walls, introducing in the analysis the effect of elastic and geometric properties of these elements and enhancing the box-like behaviour of these structures. Due to the lack of information regarding the mechanical properties of timber elements, an average value of 10.5 GPa and 7.0 GPa were assumed to the Young’s modulus related to the orthogonal and perpendicular direction of the floor warping, E1;eq and E2;eq respectively, which values are within the range suggested in [60,61,29]. Fig. 8(d) illustrates the retrofitting of wall-to-roof connections solution (S4 ), ensured by applying the same technique as in solution S1 but at the roof level, by introducing steel tie-rods underneath the ceiling joists, to sustain horizontal thrusts in the case of an earthquake. The retrofitting package RP2 adds the retrofitting solution S5 to the preceding package RP 1 , and comprehends the introduction of reinforced concrete strapping beams at the top of stone masonry walls (also designated as RC ring courses), executed along the whole perimeter of the building, enhancing the connection between roof and load-bearing stone masonry and improving both bending and shear capacity (see Fig. 10). These beams were defined with a cross-section 20 cm230 cm2 (4/10 longitudinal steel bars and /6//.30 stirrups), and classes C20/25 to concrete and S400 to steel bars, accordingly to the available structural drawings. Finally, retrofitting package RP 3 , includes the retrofitting solution S6 , which involves the shear strengthening and confinement

50

R. Maio et al. / Engineering Structures 141 (2017) 41–58

Fig. 8. Details of retrofitting solutions of retrofitting package RP 1 , adapted from Costa [58].

of masonry structural walls by the implementation of reinforced render, as specified by Costa [21]. Thus, as illustrated in Fig. 11, after the application of a first layer of filling mortar in the proportion of 1:3 (local sand extracted from Fajã Beach: Portland cement: water) for voids and surface regularisation, a 0.5 mm thick welded steel mesh made of S275 steel and 10 mm spaced ribs, is then fixed on both sides of the masonry wall through a system composed of M20 galvanised screws, /20 galvanised steel threaded rods and 4 mm thick anchor plates (20 mm  20 mm, spaced 150 cm. Finally, a 3 cm thick second layer of fine sand-blasted finishing mortar is applied [21]. This retrofitting solution was simulated in

the model by considering the reinforced plaster improving parameter for masonry, which comprises an improvement of about 2.5 times on the mechanical properties of the stone masonry typology assigned to the original BR condition of both buildings. 5. Discussion of the results The influence of the previous retrofitting packages is discussed in this section from the perspective of the seismic performancebased assessment, comparing the global indicators recommended by EN 1998-1, EN 1998–3, NTC 2008 and ATC-40 seismic codes

R. Maio et al. / Engineering Structures 141 (2017) 41–58

51

Fig. 9. Position of the tie-rods (in blue) in the Building A (left) and Building B (right). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 10. Details of retrofitting solution S5 incorporated in the retrofitting package RP2 , adapted from Costa [58].

Fig. 11. Details of retrofitting solution S6 incorporated in the retrofitting package RP3 , adapted from Costa [58].

[32,43,33] to be used on the safety verification of the ULS, which were already described in Section 3.3. Additionally, analytical fragility curves are derived by taking the target displacement of the structure, dt , as the intensity measure (IM). 5.1. Comparison in terms of seismic performance-based assessment This sub-section is divided into two different categories of results. Firstly, the results obtained from the application of the N2 Method (previously described in Section 3.3.2), referring to the Azores seismic action recommended in [45], are presented and discussed. Secondly, the results obtained from the application of the CSM by using a set of real ground motion records, as described in Section 3.3.2, are presented and compared to the previous approach. The comparison of the results obtained to each one of the mentioned global indicators, discretised by direction, lateral load pattern and building condition (from BR to RP 3 ), is provided in the following figures. In Fig. 12 it is possible to observe a quite signif-

icant dispersion of the q ratio values for the case of the BR condition, which can be explained by the lack of connection between walls intentionally induced in the numerical model. In general, it was observed that not only q decreases by implementing these retrofitting packages cumulatively, but also that the safety condition q < 3 is verified in a general manner for all the retrofitting packages, exception made to the pseudo-triangular load pattern distribution of RP1 and RP2 . From Fig. 12(a) one can conclude that Building A presents a slightly better behaviour in terms of the q ratio for the analyses computed with the uniform load pattern distribution. In the case of Building B, while in the U x direction all building conditions verify the safety requirement q < 3, in the U y direction only RP3 verifies this condition.   In what regards the ratio du =dt , depicted in Fig. 13, it is observed that only RP 3 fulfils the required safety condition   du =dt > 1, excepting for the U y direction of Building B. However, while results demonstrate some consistency in the case of Build  ing A, with a general increase of du =dt , the same does not apply in the case of Building B, particularly in the negative direction of

52

R. Maio et al. / Engineering Structures 141 (2017) 41–58

(a) Building A

(b) Building B

Fig. 12. Results of the seismic performance-based assessment in terms of the q factor.

U y , in which BR shows the better results than any other retrofitting package. Finally, the values achieved for the ratio ag;max =agR , presented in the following Fig. 14 (a), confirm the trend of the previous indica  tor du =dt , as most of the analyses in the case of the retrofitting package RP3 fulfil the required safety verification ag;max =agR > 1, exception made for the analyses carried out for Building B, in the U y direction. Moreover, and similarly to what was observed in Figs. 12 and 13, the enhancement the seismic performance is quite unclear between RP1 to RP 2 , raising some doubts about the efficiency of retrofitting solution S5 . In fact, this might be explained either due to the actual efficiency of the reinforced-concrete ring beams as retrofitting solution or to the numerical inability at simulating its influence. Additionally, as previously mentioned, the retrofitting package RP 1 resorts to retrofitting solutions from S1 to S4 , of low-to-moderate intrusiveness, which are known to be effective on enhancing the box-like behaviour of stone masonry buildings, i.e., on improving the out-of-plane behaviour of masonry structures. In this sense, since non-linear static (pushover) analyses were carried out to exclusively assess the in-plane capacity of the model, it was not expected that RP 1 could affect significantly the in-plane global capacity of the structure. In fact, several authors have been debating the actual contribution of this widely applied technique, arguing that it may lead to the development of local mechanisms in load-bearing walls when subjected to seismic loads, due to important local increments in terms of mass and stiffness at the top of masonry walls, introducing significant changes over the expected vibration modes, particularly not desirable in high-rise buildings. The results in terms of the ratio between the displacement of the MDoF system versus the ratio between the maximum admissi-

ble ground acceleration, ag;max and the reference ground acceleration, agR , shown in Fig. 15, demonstrate a global improvement of the seismic performance of both case study buildings, particularly in terms of ag;max =agR ratio. It is also possible to observe that in the case of Building A, retrofitting packages RP 1 and RP 2 allow the structure to withstand larger displacements than RP 3 . Finally, when analysing the results of the seismic performancebased assessment obtained by using the Capacity Spectrum Method for deriving capacity curves and subsequently determining the target displacement using the set of real ground motion records, as presented in Table 2 (Section 3.3.2), it is possible to observe that in general terms, the values of the ratio F m =m increase by implementing these retrofitting packages, following the trend of the N2 Method, as shown in Fig. 16. In addition, Fig. 16 shows that the target displacements values, dt , have also decreased in both case studies with the implementation of these retrofitting packages, as expected. 5.2. Comparison in terms of fragility curves The seismic performance of the modelled buildings is hereinafter compared in terms of fragility curves, analytically derived based on the spectral response approach proposed by FEMA-NIBS procedure [62], in which fragility curves describe the probability of exceeding a determined damage state, ds, for a given value of spectral displacement, Sd (herein taken as intensity measure). In this study, damage states were adopted and defined directly from yielding and ultimate displacements of the capacity curve, by means of pre-set displacement-based damage state thresholds, Sd;ds , initially developed within the framework of the RISK-EU project for several building typologies within European towns based

53

R. Maio et al. / Engineering Structures 141 (2017) 41–58

(a) Building A

(b) Building B 



Fig. 13. Results of the seismic performance-based assessment in terms of the ratio du =dt .

on expert opinion, and proposed in [63]. These damage state thresholds, that have since been widely applied in many other studies worldwide, including many examples in Portugal, such as in [15,34] for example, are defined by the following expressions: Sd;1 ¼ 0:7Sdy ; Sd;2 ¼ 1:5Sdy ; Sd;3 ¼ 0:5ðSdy þ Sdu Þ, and Sd;4 ¼ Sdu , which are associated to slight damage, moderate damage, extensive damage and near collapse damage conditions, respectively. Fragility curves were thus derived by means of lognormal functions that describe the probability of reaching or exceeding a determined damage state, ds, given the spectral displacement, Sd , according to the following Eq. (2), wherein U is the standard normal cumulative distribution function, bds is the standard deviation of the natural logarithm of spectral displacement for a given ds, and Sd;ds is the median value of the spectral displacement at which a building reaches the threshold of damage state ds.

"

1 Sd P½dsjSd  ¼ U ln bds Sd;ds

!#

ð2Þ

The parameter bds , expressed by the following Eq. (3), considers different sources of variability and uncertainty related to the model inherent characteristics and assumptions (b ), the seismic demand (bD ), the capacity curve (bC ), which comprises both geometrical and mechanical input parameters responsible for influencing the seismic global response of structures, and the definition of the damage states thresholds (bLS ).

bds ¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi b2 þ b2D þ b2C þ b2LS

ð3Þ

While the value of b was assigned equal to 0.20, according to Meireles et al. [64], bD was assumed equal to 0.25, according to Simões et al. [34], which argued that there is still a lack of information on the current EN 1998–1 [32], for a more precise estimation of this parameter. As the variability of the masonry typology of the assessed buildings might be relatively high, not only due to the quality of construction, mechanical properties of materials and constructive details, but also due to the uncertainty related to the obtained results for the seismic response, bC was assigned equal to 0.35, accordingly to some of the most recent studies found in literature concerning this parameter [65,64,34]. The value of bLS was determined assuming that the displacement limit thresholds, Sd;ds , correspond to the conditional probability of 50% of reaching or exceeding the corresponding damage state threshold. Thus, by assuming a uniform probability density function (in an interval around Sd;ds ), as proposed by Lagomarsino and Cattari [66] and Pagnini et al. [65], the resulting value of bLS varies, for each building and type of analysis, as a function of the ductility of the capacity curve. Table 4 summarises the results obtained for each retrofitting condition of both case study buildings, namely the target displacements, dt , obtained for the elastic response spectra recommended in EN 1998–1 [32], the yielding and ultimate displacements, dy and du , ductility, l, and the bds values. It is worth referring that in order to compare the global influence of the considered retrofitting solutions and to reduce the influence of a particular analysis on fragility curves’ derivation process, the mean values of the yielding and ultimate displacements, dy and du , as well as the

54

R. Maio et al. / Engineering Structures 141 (2017) 41–58

(a) Building A

(b) Building B

Fig. 14. Results of the seismic performance-based assessment in terms of ratio ag;max =agR .

(a) Building A

(b) Building B

Fig. 15. Results of the seismic performance-based assessment in terms of the relationship between the displacement of the MDoF system versus the ratio between the maximum admissible ground acceleration, ag;max and the reference ground acceleration, agR .

target displacements, dt , were assumed from a total of 84 analyses for each direction and retrofitting condition. Moreover, the respective fragility curves derived in function of the spectral displacement, Sd (in this case dt ), are given in Fig. 17, where the vertical lines represent the mean target displacements obtained following the recommendations of the EN 1998-1 [32] (in red), and the mean target displacement obtained through the CSM Method to the set of real ground motion records (short dots in blue). Regarding the latter, the filled area (in light blue) is limited by the corresponding minimum and maximum target displace-

ment values of this set of records. However, it is worth referring that Rec. 6 was not included on the estimation of this mean target displacement, due to the bias associated with this particular record, previously discussed in Section 3.3.2. When comparing the results from Fig. 17, one can observe that there is an overall enhancement of the seismic performance of both case study buildings, as the expected damage decreases with the implementation of these cumulative retrofitting packages. This improvement is particularly remarkable in the case of the mean target displacement obtained through the CSM Method, as

55

R. Maio et al. / Engineering Structures 141 (2017) 41–58

(a) Building A

(b) Building B 

Fig. 16. Results of the seismic performance-based assessment in terms of the target displacement, dt , for all the building conditions.

Table 4 Results obtained for each retrofitting condition by applying the N2 Method. Building A

dy [cm]

du [cm]

l [–]

b1 [–]

b2 [–]

b3 [–]

b4 [–]

Ux Uy Ux Uy Ux Uy Ux Uy

1.41 1.55 1.49 1.72 1.37 1.62 0.88 1.13

0.28 0.26 0.57 0.55 0.52 0.61 0.36 0.59

1.10 0.62 2.76 2.00 2.31 2.08 2.12 2.97

3.89 2.41 4.83 3.63 4.42 3.42 5.84 5.01

0.541 0.541 0.541 0.541 0.541 0.541 0.541 0.541

0.513 0.488 0.526 0.510 0.520 0.507 0.544 0.529

0.494 0.484 0.501 0.492 0.498 0.490 0.508 0.503

0.487 0.482 0.488 0.486 0.488 0.485 0.490 0.489

Ux Uy Ux Uy Ux Uy Ux Uy

1.37 1.30 1.57 1.39 1.59 1.39 0.94 0.77

0.30 0.30 0.83 0.34 0.91 0.35 0.89 0.28

1.07 0.76 2.92 0.84 2.63 1.43 4.18 1.40

3.60 2.52 3.49 2.49 2.89 4.04 4.70 5.04

0.541 0.541 0.541 0.541 0.541 0.541 0.541 0.541

0.509 0.490 0.508 0.490 0.497 0.515 0.524 0.530

0.492 0.485 0.491 0.484 0.487 0.495 0.500 0.503

0.486 0.482 0.486 0.482 0.484 0.487 0.488 0.489

Direction

BR RP1 RP2 RP3

B

dt [cm]

Retrofitting condition

BR RP1 RP2 RP3

globally, the use of the elastic response spectra as seismic demand through applying the N2 Method leads, as expected, to more conservative results. In the case of retrofitting package RP 3 , this difference was found even more significant, particularly in the U x direction, as target displacements obtained for the elastic response spectra recommended by the EN 1998–1 (in Table 4) exceed the maximum target displacement values obtained by considering a set of real ground motion records (short dots in blue in Fig. 17). However, in the BR condition, this tendency appears to be reversed. It is important noting that all the studied retrofitting packages reduces the dispersion of the structural performance point results due to the aleatory uncertainty inherent to the real earthquake ground motions (most evident for the RP3 case), which leads to a reduction of the seismic risk. It is also worth referring that, similarly to what was observed in Section 5.1, as retrofitting packages are implemented, the resulting target displacement decreases for both types of seismic demand herein considered, leading to lower levels of expected damage. However, once again, no significant improvements were observed in the case of retrofitting package RP 2 . 6. Conclusions This paper approaches the seismic response of a traditional stone masonry building assessed by means of non-linear static (pushover) analysis resorting to the commercial version of 3MuriÒ software. The authors aimed not only to bring attention to the high

vulnerability generally associated to these type of structures, but also to perceive the influence of different sustainable, cost-wise and widely applied retrofitting solutions over the structure’s inplane seismic response. With respect to the BR condition, the authors believe that the need to disregard the presence of the external stone masonry stairs, due to modelling issues, may have exacerbated these analyses. Moreover, further analysis of local mechanisms should be addressed in order to evaluate the real contribution of retrofitting package RP1 over the structure’s out-of-plane capacity and therefore assess the overall benefit of such retrofitting solutions. In order understand the real efficiency of reinforced concrete strapping beams solutions (S5 ), the authors believe that further analysis resorting to both numerical and experimental approaches should be performed, particularly when referring to multi-storeys stone masonry buildings. Despite the foregoing, from the global seismic performance-based assessment analyses carried out, the retrofitting package RP3 was able to fulfil the ultimate limit states’ requirements of current seismic codes (EN 1998–1, EN 1998–3 and Italian code NTC 2008) for the majority of the performed analyses, except for a few analysis in the U y direction of Building B, confirming the pronounced influence of the masonry quality over the in-plane capacity of structures. It is worth referring that the criteria herein adopted for the selection of the control node and the worst case scenario analyses might have exacerbated the results, as they corresponded in practically all cases to the pseudotriangular load pattern distribution, which the representativeness

56

R. Maio et al. / Engineering Structures 141 (2017) 41–58

(a) Building A

(b) Building B

Fig. 17. Fragility curves for both directions, with the representation of the target displacements obtained by applying the N2 Method (in red) and the mean of the target displacements obtained by applying the CSM Method (short dots in blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

in the case of old masonry structures is still an open issue [67]. Resorting to real ground motion records as seismic demand for determining the target displacements of existing structures allowed to quantitatively compare the outputs obtained by using the recommended code elastic response spectra. Even though in this work different sources of uncertainty were considered for deriving analytical fragility curves, further research should be conducted to better estimate uncertainties in numerical models. Moreover, the authors will engage into sensitivity analysis in future works, in order to explicitly address epistemic uncertainties and subsequently enhance the reliability of seismic performancebased assessment procedures. Acknowledgements This work was supported by the Portuguese Foundation of Science and Technology (FCT) through a PhD scholarship of the Doctoral Program Analysis and Mitigation of Risks in Infrastructures (InfraRisk-). The authors acknowledge to the Society of Promotion for Housing and Infrastructures Rehabilitation (SPRHI) and to the Regional Secretariat for Housing and Equipment (SRHE)

of Faial for their support and contribution to the development of this work. Last but not least, we would like to express our gratitude to the reviewers for their insightful comments on the paper, as these comments led us to a substantial improvement of the work herein presented.

References [1] Silva V, Marques M, Castro JM, Varum H. Development and application of a real-time loss estimation framework for portugal. Bull Earthquake Eng 2015;13(9):2493–516. http://dx.doi.org/10.1007/s10518-015-9743-0. [2] Moreira V. Earthquakes at north-atlantic passive margins: neotectonics and postglacial rebound. Seismicity of the Portuguese Continental Margin, nato asi series Edition. Netherlands, Dordrecht: Springer; 1989. p. 533–45. Ch. 266. [3] Nunes JC. General description of the Azores region seismicity. In: Oliveira C, Costa A, Nunes J, editors. Sismo 1998 – Açores. Uma década depois, Sersilito empresa gráfica, Lda, Horta, Faial Island, Azores, Portugal. p. 59–72. Chapter 1. [4] Cabral J, Moniz C, Batilo J, Figueiredo P, Carvalho J, Matias L, Teves-Costa P, Dias R, Simão N. The 1909 Benavente (Portugal) earthquake: search for the source. Nat Hazards 2013;69(2):1211–27. [5] Choffat P. Le tremblement de terre du 23 Avril 1909 dans le Ribatejo, Revista de Obras Públicas e Minas Tomo XLIII. [6] Matias L, Dias NA, Morais I, Vales D, Carrilho F, Madeira J, Gaspar JL, Senos L, Silveira AB. The 9th of July 1998 Faial Island (Azores, North Atlantic) seismic

R. Maio et al. / Engineering Structures 141 (2017) 41–58

[7] [8]

[9]

[10] [11]

[12] [13]

[14]

[15]

[16]

[17]

[18] [19]

[20]

[21]

[22]

[23] [24]

[25]

[26] [27]

[28]

[29] [30]

[31]

[32]

[33]

sequence. J Seismol 2007;11(3):275–98. http://dx.doi.org/10.1007/s10950007-9052-4. Oliveira CS, Costa A, Nunes JC, editors. Sismo 1998 – Açores. Uma década depois. Lda: Sersilito empresa gráfica; 2008. Ferreira T. Out-of-plane seismic performance of stone masonry walls: experimental and analytical assessment Phd thesis. Department of Civil Engineering of the University of Aveiro; 2015. Neves F, Costa A, Vicente R, Oliveira CS, Varum H. Seismic vulnerability assessment and characterization of the buildings on Faial Island, Azores. Bull Earthquake Eng 2012;10(1):27–44. http://dx.doi.org/10.1007/s10518-0119276-0. O.D. Cardona, A.H. Barbat, El riesgo Sísmico y su prevención, Madrid and Calidad Siderúrgica, 2000. Silva V, Crowley H, Varum H, Pinho R. Seismic risk assessment for mainland Portugal. Bull Earthquake Eng 2014. http://dx.doi.org/10.1007/s10518-0149630-0. Maio R. Seismic vulnerability assessment of old building aggregates Msc thesis. Department of Civil Engineering of the University of Aveiro; 2013. Maio R, Vicente R, Formisano A, Varum H, Maio R, Vicente R, Formisano A, Varum H. Seismic vulnerability of building aggregates through hybrid and indirect assessment techniques. Bull Earthquake Eng 2015. http://dx.doi.org/ 10.1007/s10518-015-9747-9. Santos C, Ferreira TM, Vicente R, Mendes da Silva JAR. Building typologies identification to support risk mitigation at the urban scale – Case study of the old city centre of Seixal, Portugal. J Cult Heritage 2013;14(6):449–63. http:// dx.doi.org/10.1016/j.culher.2012.11.001. Vicente R, Parodi S, Lagomarsino S, Varum H, Mendes da Silva JAR. Seismic vulnerability and risk assessment: case study of the historic city centre of Coimbra, Portugal. Bull Earthquake Eng 2011;9(4):1067–96. http://dx.doi.org/ 10.1007/s10518-010-9233-3. Portugal, Ministerial Order No. 53/2014, of April 8, Ministry for Environment and Spatial Planning, Official Gazzette of the Republic of Portugal, 1st series, No. 69 (in Portuguese); 2014. Maio R, Ferreira TM, Vicente R, Estêvão JMC. Seismic vulnerability assessment of historical centres: case study of the old city centre of Faro, Portugal. J Risk Res 2015. http://dx.doi.org/10.1080/13669877.2014.988285. Neves F. Avaliação da vulnerabilidade sísmica do parque habitacional da Ilha do Faial Msc thesis. University of Aveiro; 2008. Indirli M, Kouris LAS, Formisano A, Borg RP, Mazzolani FM. Seismic damage assessment of unreinforced masonry structures after the Abruzzo 2009 earthquake: the case study of the historical centers of L’Aquila and Castelvecchio Subequo. Int J Archit Heritage 2013;7(5):536–78. http://dx.doi. org/10.1080/15583058.2011.654050. Gautam D, Bhetwal KK, Rodrigues H, Neupane P, Sanada Y. Observed damage patterns on buildings during 2015 gorkha (nepal) earthquake. Proceedings of the 14th International Symposium on New Technologies for Urban Safety of Mega Cities in Asia, no. 14, Kathmandu, Nepal. 2015. p. 8. Costa A. Determination of mechanical properties of traditional masonry walls in dwellings of Faial Island, Azores. Earthquake Eng Struc Dyn 2002;31 (February 2000):1361–82. http://dx.doi.org/10.1002/eqe.167. Ferreira TM, Costa AA, Arêde A, Varum H, Costa A. In situ out-of-plane cyclic testing of original and strengthened traditional stone masonry walls using airbags. J Earthquake Eng 2016;20(5):749–72. http://dx.doi.org/10.1080/ 13632469.2015.1107662. Costa A, Arêde A. Strengthening of structures damaged by the Azores earthquake of 1998. Constr Build Mater 2006;20(4):252–68. Costa A, Arêde A, Costa AA, Oliveira CS. In situ cyclic tests on existing stone masonry walls and strengthening solutions. Earthquake Eng Struct Dyn 2011;40:449–71. http://dx.doi.org/10.1002/eqe.1046. Costa AA. Seismic assessment of the out-of-plane performance of traditional stone masonry walls Phd thesis. Faculty of Engineering of the University of Oporto; 2012. Neves N, Arêde A, Costa A. Seismic analysis of a building block. Bull Earthquake Eng 2012:1–33. http://dx.doi.org/10.1007/s10518-011-9310-2. Ferreira TM, Maio R, Vicente R, Costa A. Earthquake risk mitigation: the impact of seismic retrofitting strategies on urban resilience. Int J Strategic Property Manage 2016;20(3):291–304. http://dx.doi.org/10.3846/ 1648715X.2016.1187682. Ferreira T, Maio R, Vicente R, Ferreira T, Maio R, Vicente R. Seismic vulnerability assessment of the old city centre of Horta, Azores: calibration and application of a vulnerability index method. Bull Earthquake Eng 2016. http://dx.doi.org/10.1007/s10518-016-0071-9. STADATA, 3Muri program, Release 10.0; 2011. Lagomarsino S, Penna A, Galasco A, Cattari S. TREMURI program: an equivalent frame model for the nonlinear seismic analysis of masonry buildings. Eng Struct 2013;56:1787–99. http://dx.doi.org/10.1016/j.engstruct.2013.08.002. Penna A, Lagomarsino S, Galasco A. A nonlinear macroelement model for the seismic analysis of masonry buildings. Earthquake Eng Struct Dyn 2014;43:159–79. http://dx.doi.org/10.1002/eqe.2335. CEN, Eurocode 8: design of structures for earthquake resistance – part I: general rules, seismic actions and rules for buildings (EC8-1). European Committee for Standardization, Brussels, Belgium; 2004. Italy, NTC 2008, Decreto Ministeriale 14/1/2008, Norme tecniche per le costruzioni. Ministry of Infrastructures and Transportations, Gazzetta Ufficiale S.O. no.30 on 4/2/2008; 2008.

57

[34] Simões A, Miloševic´ J, Meireles H, Bento R, Cattari S, Lagomarsino S. Fragility curves for old masonry building types in Lisbon. Bull Earthquake Eng 2015. http://dx.doi.org/10.1007/s10518-015-9750-1. [35] Turnšek V, Sheppard P. The shear and flexural resistance of masonry walls. International Research Conference on Earthquake Engineering, Skopje, Japan.1980. p. 517–73. [36] Cattari S, Lagomarsino S. Seismic assessment of mixed masonry-reinforced concrete buildings by non-linear static analyses. Earthquakes Struct 2013;4 (3):241–64. [37] Giongo I, Wilson A, Dmytro Y, Derakhshan H, Tomasi H, Griffith M, Quenneville P, Ingham J. Detailed seismic assessment and improvement procedure for vintage flexible timber diaphragms. Bull N Z Soc Earthquake Eng 2014;47 (2):97–118. [38] NZSEE, Assessment and improvement of the structural performance of buildings in earthquake. recommendations of a nzsee study group, New Zealand Society for Earthquake Engineering, Wellington, New Zealand; 2015. [39] ASCE. ASCE/SEI. Seismic evaluation and retrofit of existing buildings. Reston, VA: American Society of Civil Engineers; 2014. [40] RSAEEP, Regulamento de segurança e acções para estruturas de edifícios e pontes, Decreto-Lei no. 253/83 de 31 de Maio; 1983. [41] Maio R, Tsionis G. Seismic fragility curves for the European building stock: Review and evaluation of analytical fragility curves, Tech. rep., Joint Research Centre, EUR 27635 EN; 2016. doi:10.2788/586263. [42] Simões A, Bento R, Cattari S, Lagomarsino S. Seismic performance-based assessment of Gaioleiro buildings. Eng Struct 2014;80:486–500. http://dx.doi. org/10.1016/j.engstruct.2014.09.025. [43] CEN, Eurocode 8: design of structures for earthquake resistance – part 3: assessment and retrofitting of buildings (EC8-3). European Committee for Standardization, Brussels, Belgium; 2004. [44] Fajfar P. Capacity spectrum method based on inelastic demand spectra. Earthquake Eng Struc Dyn 1999;28(9):979–93. http://dx.doi.org/10.1002/ (SICI)1096-9845(199909)28:9979:AIDEQE8503.0.CO. [45] CEN, Eurocódigo 8: projecto de estruturas para resistência aos sismos - parte 1: Regras gerais, ações sísmicas e regras para edifícios (EC8-1). Norma Portuguesa. NP EN 1998–1 2010. European Committee for Standardization, Brussels, Belgium; 2010. [46] ATC-40, Seismic Evaluation and Retrofit of Concrete Buildings. Report: SSC 96– 01, Volume 1, Tech. rep., Applied Technology Council, Redwood City, California; 1996. [47] Oliveira CS. Strong motion records obtained in the accelerometric network of Azores. Comparison with other indicators of the seismic movement. Sismo 1998 – Açores. Uma década depois; 2008. p. 119–36. [48] Estêvão JMC, Oliveira CS. Point and fault rupture stochastic methods for generating simulated accelerograms considering soil effects for structural analysis. Soil Dyn Earthquake Eng 2012;43:329–41. http://dx.doi.org/10.1016/ j.soildyn.2012.07.019. [49] Teves Costa P, Pacheco J, Escuer M, Team CT. The use of ambient vibrations to estimate the dynamic characteristics of surface layers of the town of Horta, Faial. In: Oliveira C, Costa A, Nunes J, editors. Sismo 1998 – Açores. Uma década depois. Lda, Horta, Faial Island, Azores, Portugal: Sersilito empresa gráfica; 2008. p. 137–49. Chapter 9. [50] Estêvão JMC, Carvalho A. The role of source and site effects on structural failures due to Azores earthquakes. Eng Failure Anal 2015. http://dx.doi.org/ 10.1016/j.engfailanal.2014.12.010. [51] Formisano A, Marzo A. Simplified and refined methods for seismic vulnerability assessment and retrofitting of an italian cultural heritage masonry building. Comput Struct 2017;180:13–26. http://dx.doi.org/ 10.1016/j.compstruc.2016.07.005. cIVIL–COMP 2. [52] Faggiano B, Marzo A, Formisano A, Mazzolani FM. Review: innovative steel connections for the retrofit of timber floors in ancient buildings: a numerical investigation. Comput Struct 2009;87(1–2):1–13. http://dx.doi.org/10.1016/ j.compstruc.2008.07.005. [53] Brignola A, Pampanin S, Podestà S. Evaluation and control of the in-plane stiffness of timber floors for the performance-based retrofit of URM buildings. Bull N Z Soc Earthquake Eng 2009;42(3):204–21. [54] Araújo AS, Oliveira DV, Lourenço PB. Numerical study on the performance of improved masonry-to-timber connections in traditional masonry buildings. Eng Struct 2014;80:501–13. http://dx.doi.org/10.1016/j.engstruct.2014.09. 027. [55] Moreira S, Ramos LF, Oliveira DV, Lourenço PB. Design parameters for seismically retrofitted masonry-to-timber connections: injection anchors. Int J Archit Heritage 2016;10(2–3):217–34. http://dx.doi.org/10.1080/15583058. 2015.1113339. [56] Carvalho EC, Oliveira CS, Fragoso M, Miranda V. Regras gerais de reabilitação e reconstrução de edifícios correntes afectados pela crise s´smica do Faial, Pico e São Jorge, iniciada pelo sismo de 9 de Julho de 1998, Tech. rep., LREC/DMCE REl. 100/98; 1998. [57] Oliveira CS, Lucas A, Guedes JHC, MONOGRAFIA – 10 Anos após o Sismo dos Açores de 1 de Janeiro de 1980 - Volume I e II, Secretaria Regional da Habitação e Obras Públicas, Delegação da Ilha Terceira, Açores, Laboratório Nacional de Engenharia Civil, Angra do Herósmo, Açores, Portugal; 1990. [58] Costa A. Strengthening and repairing earthquake damaged structures. In: Oliveira C, Roca A, Goula X, editors. Assessing and managing earthquake risk. Netherlands, Dordrecht, Netherlands: Springer; 2006. p. 403–26. Ch. 19.

58

R. Maio et al. / Engineering Structures 141 (2017) 41–58

[59] D’Ayala D, Speranza E, Housing Report: Single-family stone masonry house, Tech. rep., Earthquake Engineering Research Institute (EERI) and International Association for Earthquake Engineering (IAEE), Italy; 2002. [60] Dias T. Pavimentos de madeira em edifícios antigos: diagnóstico e intervenção estrutural, Mestrado em reabilitação do património edificado. Faculty of Engineering of the University of Oporto; 2008. [61] Fiorucci S. Ristrutturazione e riabilitazione strutturale del castello di sant’apollinare a marsciano (PG) interessato dal sisma del 15 dicembre 2009 Bachelor thesis (tesi di laurea). Università degli Studi Guglielmo Marconi; 2014. [62] FEMA, Earthquake loss estimation methodology technical report, HAZUS 99, Tech. rep., HAZUS 99 Service Release 2, prepared by the National Institute of Building Sciences for the Federal Emergency Management Agency, Washington DC, USA; 2001.

[63] Lagomarsino S, Giovinazzi S. Macroseismic and Mechanical Models for the Vulnerability and Damage Assessment of Current Buildings. Bull Earthquake Eng 2006;4(4):415–43. http://dx.doi.org/10.1007/s10518-006-9024-z. [64] Meireles H, Bento R, Cattari S, Lagomarsino S. Seismic assessment and retrofitting of Pombalino buildings by pushover analyses. Earthquake Struct 2014;7(1):57–82. http://dx.doi.org/10.12989/eas.2014.7.1.057. [65] Pagnini LC, Vicente R, Lagomarsino S, Varum H. A mechanical model for the seismic vulnerability assessment of old masonry buildings. Earthquake Struct 2011;2(1):25–42. http://technopress.kaist.ac.kr/?journal=eas&subpage=5. [66] Lagomarsino S, Cattari S. PERPETUATE guidelines for seismic performancebased assessment of cultural heritage masonry structures. Bull Earthquake Eng 2014;13(1):13–47. http://dx.doi.org/10.1007/s10518-014-9674-1. [67] Lourenço PB, Mendes N, Ramos LF, Oliveira DV. Analysis of masonry structures without box behavior. Int J Archit Heritage 2011;5(4–5):369–82. http://dx.doi. org/10.1080/15583058.2010.528824.